AIAssistantsModelConfig
Configuration for a single model tier (embedding or reasoning). On write, supply model_id, api_url, and api_key (when required). On read, all fields except api_key are returned.
{
"model_id": "string",
"model_name": "string",
"sdk_provider": "string",
"api_url": "string",
"api_key": "string"
}
Stable unique identifier for this model (e.g. openai_gpt-4.1). Not intended for display in the UI.
Raw API model name submitted to the model provider's API (e.g. gpt-4.1). Resolved by the backend from model_id. Not accepted in write requests.
SDK provider identifier (e.g. google, openai). Resolved by the backend from model_id. Not accepted in write requests.
Endpoint URL for this model tier. Supplied by the operator.
Credential for authenticating to the model endpoint. Supplied by the operator. Required for non-GCP tiers. Optional for GCP-backed embedding tiers when GCP service-account authentication is used. Never returned in read responses.